Add uint128

[aritkulova]

No description provided.

[LesterEvSe]

Okay, I see the problem with running the tests, they take a lot of time. I have been playing around with the code, and the main problem is that with match if_test_this_function(...) we run a lot of similar operations, and in fact, to reach, for example, assert_bool(is_zero_128(a), expected_bool);, we need to go through 10 checks beforehand.

For example, when I run code like this, simplex tests finishes in under 3s, compared to the current 45s:

fn main() {
    let fn_idx: u8 = witness::FUNCTION_INDEX;

    let a: u128 = witness::FIRST_ARG;
    let b: u128 = witness::SECOND_ARG;
    let expected: Option<u128> = witness::EXPECTED;

    let expected_bool: bool = witness::EXPECTED_BOOL;
    let second_expected: u128 = witness::SECOND_EXPECTED; // for cases where the result is a 256 bit value
    let third_expected: u128 = witness::THIRD_EXPECTED; // for cases where the result is 256 bit + 128 bit values

    match if_test_this_function(197, fn_idx) { true => { assert_bool(is_zero_128(a), expected_bool); }, false => (), };
}

Note: The same optimization is true for other operations as well. I checked this for fn_idx == 195.

I have two possible solutions:

• We could split u128_test.simf into several smaller files. I think this would partially fix the problem of tests taking too long to run.
• Another solution is to wait for feat: add enum declarations and multi-arm enum match expressions SimplicityHL#336 to be merged. Then we would have a multi-arm match block with an enum, and we could probably solve this problem that way too.

[LesterEvSe]

Please add tests for this optimization. Suggested cases: a_low == b_low, a_low - b_low == u64::MAX, and if possible a property test asserting q * b + r == a on randomized inputs

[LesterEvSe]

ACK b001da3; tested locally with simplex test --test-threads 10

[Hrom131]

LGTM!